clang -cc1 -cc1 -triple amd64-unknown-openbsd7.0 -analyze -disable-free -disable-llvm-verifier -discard-value-names -main-file-name ELFObjectFile.cpp -analyzer-store=region -analyzer-opt-analyze-nested-blocks -analyzer-checker=core -analyzer-checker=apiModeling -analyzer-checker=unix -analyzer-checker=deadcode -analyzer-checker=cplusplus -analyzer-checker=security.insecureAPI.UncheckedReturn -analyzer-checker=security.insecureAPI.getpw -analyzer-checker=security.insecureAPI.gets -analyzer-checker=security.insecureAPI.mktemp -analyzer-checker=security.insecureAPI.mkstemp -analyzer-checker=security.insecureAPI.vfork -analyzer-checker=nullability.NullPassedToNonnull -analyzer-checker=nullability.NullReturnedFromNonnull -analyzer-output plist -w -setup-static-analyzer -mrelocation-model static -mframe-pointer=all -relaxed-aliasing -fno-rounding-math -mconstructor-aliases -munwind-tables -target-cpu x86-64 -tune-cpu generic -debugger-tuning=gdb -fcoverage-compilation-dir=/usr/src/gnu/usr.bin/clang/libLLVM/obj -resource-dir /usr/local/lib/clang/13.0.0 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/AMDGPU -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Analysis -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ASMParser -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/BinaryFormat -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Bitcode -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Bitcode -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Bitstream -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /include/llvm/CodeGen -I /include/llvm/CodeGen/PBQP -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/IR -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/IR -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/Coroutines -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ProfileData/Coverage -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/CodeView -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/DWARF -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/MSF -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/PDB -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Demangle -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ExecutionEngine -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ExecutionEngine/JITLink -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ExecutionEngine/Orc -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Frontend -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Frontend/OpenACC -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Frontend -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Frontend/OpenMP -I /include/llvm/CodeGen/GlobalISel -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/IRReader -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/InstCombine -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/Transforms/InstCombine -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/LTO -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Linker -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/MC -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/MC/MCParser -I /include/llvm/CodeGen/MIRParser -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Object -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Option -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Passes -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ProfileData -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/Scalar -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/ADT -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Support -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/DebugInfo/Symbolize -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Target -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/Utils -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/Vectorize -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include/llvm/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Target/X86 -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include/llvm/Transforms/IPO -I /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/include -I /usr/src/gnu/usr.bin/clang/libLLVM/../include -I /usr/src/gnu/usr.bin/clang/libLLVM/obj -I /usr/src/gnu/usr.bin/clang/libLLVM/obj/../include -D NDEBUG -D __STDC_LIMIT_MACROS -D __STDC_CONSTANT_MACROS -D __STDC_FORMAT_MACROS -D LLVM_PREFIX="/usr" -internal-isystem /usr/include/c++/v1 -internal-isystem /usr/local/lib/clang/13.0.0/include -internal-externc-isystem /usr/include -O2 -Wno-unused-parameter -Wwrite-strings -Wno-missing-field-initializers -Wno-long-long -Wno-comment -std=c++14 -fdeprecated-macro -fdebug-compilation-dir=/usr/src/gnu/usr.bin/clang/libLLVM/obj -ferror-limit 19 -fvisibility-inlines-hidden -fwrapv -stack-protector 2 -fno-rtti -fgnuc-version=4.2.1 -vectorize-loops -vectorize-slp -fno-builtin-malloc -fno-builtin-calloc -fno-builtin-realloc -fno-builtin-valloc -fno-builtin-free -fno-builtin-strdup -fno-builtin-strndup -analyzer-output=html -faddrsig -D__GCC_HAVE_DWARF2_CFI_ASM=1 -o /home/ben/Projects/vmm/scan-build/2022-01-12-194120-40624-1 -x c++ /usr/src/gnu/usr.bin/clang/libLLVM/../../../llvm/llvm/lib/Object/ELFObjectFile.cpp
1 | |
2 | |
3 | |
4 | |
5 | |
6 | |
7 | |
8 | |
9 | |
10 | |
11 | |
12 | |
13 | #include "llvm/Object/ELFObjectFile.h" |
14 | #include "llvm/ADT/Triple.h" |
15 | #include "llvm/BinaryFormat/ELF.h" |
16 | #include "llvm/MC/MCInstrAnalysis.h" |
17 | #include "llvm/MC/SubtargetFeature.h" |
18 | #include "llvm/Object/ELF.h" |
19 | #include "llvm/Object/ELFTypes.h" |
20 | #include "llvm/Object/Error.h" |
21 | #include "llvm/Support/ARMAttributeParser.h" |
22 | #include "llvm/Support/ARMBuildAttributes.h" |
23 | #include "llvm/Support/Endian.h" |
24 | #include "llvm/Support/ErrorHandling.h" |
25 | #include "llvm/Support/MathExtras.h" |
26 | #include "llvm/Support/RISCVAttributeParser.h" |
27 | #include "llvm/Support/RISCVAttributes.h" |
28 | #include "llvm/Support/TargetRegistry.h" |
29 | #include <algorithm> |
30 | #include <cstddef> |
31 | #include <cstdint> |
32 | #include <memory> |
33 | #include <string> |
34 | #include <system_error> |
35 | #include <utility> |
36 | |
37 | using namespace llvm; |
38 | using namespace object; |
39 | |
40 | const EnumEntry<unsigned> llvm::object::ElfSymbolTypes[NumElfSymbolTypes] = { |
41 | {"None", "NOTYPE", ELF::STT_NOTYPE}, |
42 | {"Object", "OBJECT", ELF::STT_OBJECT}, |
43 | {"Function", "FUNC", ELF::STT_FUNC}, |
44 | {"Section", "SECTION", ELF::STT_SECTION}, |
45 | {"File", "FILE", ELF::STT_FILE}, |
46 | {"Common", "COMMON", ELF::STT_COMMON}, |
47 | {"TLS", "TLS", ELF::STT_TLS}, |
48 | {"Unknown", "<unknown>: 7", 7}, |
49 | {"Unknown", "<unknown>: 8", 8}, |
50 | {"Unknown", "<unknown>: 9", 9}, |
51 | {"GNU_IFunc", "IFUNC", ELF::STT_GNU_IFUNC}, |
52 | {"OS Specific", "<OS specific>: 11", 11}, |
53 | {"OS Specific", "<OS specific>: 12", 12}, |
54 | {"Proc Specific", "<processor specific>: 13", 13}, |
55 | {"Proc Specific", "<processor specific>: 14", 14}, |
56 | {"Proc Specific", "<processor specific>: 15", 15} |
57 | }; |
58 | |
59 | ELFObjectFileBase::ELFObjectFileBase(unsigned int Type, MemoryBufferRef Source) |
60 | : ObjectFile(Type, Source) {} |
61 | |
62 | template <class ELFT> |
63 | static Expected<std::unique_ptr<ELFObjectFile<ELFT>>> |
64 | createPtr(MemoryBufferRef Object, bool InitContent) { |
65 | auto Ret = ELFObjectFile<ELFT>::create(Object, InitContent); |
66 | if (Error E = Ret.takeError()) |
67 | return std::move(E); |
68 | return std::make_unique<ELFObjectFile<ELFT>>(std::move(*Ret)); |
69 | } |
70 | |
71 | Expected<std::unique_ptr<ObjectFile>> |
72 | ObjectFile::createELFObjectFile(MemoryBufferRef Obj, bool InitContent) { |
73 | std::pair<unsigned char, unsigned char> Ident = |
74 | getElfArchType(Obj.getBuffer()); |
75 | std::size_t MaxAlignment = |
76 | 1ULL << countTrailingZeros( |
| 1 | Calling 'countTrailingZeros<unsigned long>' | |
|
| 8 | | Returning from 'countTrailingZeros<unsigned long>' | |
|
| 9 | | The result of the left shift is undefined due to shifting by '64', which is greater or equal to the width of type 'unsigned long long' |
|
77 | reinterpret_cast<uintptr_t>(Obj.getBufferStart())); |
78 | |
79 | if (MaxAlignment < 2) |
80 | return createError("Insufficient alignment"); |
81 | |
82 | if (Ident.first == ELF::ELFCLASS32) { |
83 | if (Ident.second == ELF::ELFDATA2LSB) |
84 | return createPtr<ELF32LE>(Obj, InitContent); |
85 | else if (Ident.second == ELF::ELFDATA2MSB) |
86 | return createPtr<ELF32BE>(Obj, InitContent); |
87 | else |
88 | return createError("Invalid ELF data"); |
89 | } else if (Ident.first == ELF::ELFCLASS64) { |
90 | if (Ident.second == ELF::ELFDATA2LSB) |
91 | return createPtr<ELF64LE>(Obj, InitContent); |
92 | else if (Ident.second == ELF::ELFDATA2MSB) |
93 | return createPtr<ELF64BE>(Obj, InitContent); |
94 | else |
95 | return createError("Invalid ELF data"); |
96 | } |
97 | return createError("Invalid ELF class"); |
98 | } |
99 | |
100 | SubtargetFeatures ELFObjectFileBase::getMIPSFeatures() const { |
101 | SubtargetFeatures Features; |
102 | unsigned PlatformFlags = getPlatformFlags(); |
103 | |
104 | switch (PlatformFlags & ELF::EF_MIPS_ARCH) { |
105 | case ELF::EF_MIPS_ARCH_1: |
106 | break; |
107 | case ELF::EF_MIPS_ARCH_2: |
108 | Features.AddFeature("mips2"); |
109 | break; |
110 | case ELF::EF_MIPS_ARCH_3: |
111 | Features.AddFeature("mips3"); |
112 | break; |
113 | case ELF::EF_MIPS_ARCH_4: |
114 | Features.AddFeature("mips4"); |
115 | break; |
116 | case ELF::EF_MIPS_ARCH_5: |
117 | Features.AddFeature("mips5"); |
118 | break; |
119 | case ELF::EF_MIPS_ARCH_32: |
120 | Features.AddFeature("mips32"); |
121 | break; |
122 | case ELF::EF_MIPS_ARCH_64: |
123 | Features.AddFeature("mips64"); |
124 | break; |
125 | case ELF::EF_MIPS_ARCH_32R2: |
126 | Features.AddFeature("mips32r2"); |
127 | break; |
128 | case ELF::EF_MIPS_ARCH_64R2: |
129 | Features.AddFeature("mips64r2"); |
130 | break; |
131 | case ELF::EF_MIPS_ARCH_32R6: |
132 | Features.AddFeature("mips32r6"); |
133 | break; |
134 | case ELF::EF_MIPS_ARCH_64R6: |
135 | Features.AddFeature("mips64r6"); |
136 | break; |
137 | default: |
138 | llvm_unreachable("Unknown EF_MIPS_ARCH value"); |
139 | } |
140 | |
141 | switch (PlatformFlags & ELF::EF_MIPS_MACH) { |
142 | case ELF::EF_MIPS_MACH_NONE: |
143 | |
144 | break; |
145 | case ELF::EF_MIPS_MACH_OCTEON: |
146 | Features.AddFeature("cnmips"); |
147 | break; |
148 | default: |
149 | llvm_unreachable("Unknown EF_MIPS_ARCH value"); |
150 | } |
151 | |
152 | if (PlatformFlags & ELF::EF_MIPS_ARCH_ASE_M16) |
153 | Features.AddFeature("mips16"); |
154 | if (PlatformFlags & ELF::EF_MIPS_MICROMIPS) |
155 | Features.AddFeature("micromips"); |
156 | |
157 | return Features; |
158 | } |
159 | |
160 | SubtargetFeatures ELFObjectFileBase::getARMFeatures() const { |
161 | SubtargetFeatures Features; |
162 | ARMAttributeParser Attributes; |
163 | if (Error E = getBuildAttributes(Attributes)) { |
164 | consumeError(std::move(E)); |
165 | return SubtargetFeatures(); |
166 | } |
167 | |
168 | |
169 | bool isV7 = false; |
170 | Optional<unsigned> Attr = |
171 | Attributes.getAttributeValue(ARMBuildAttrs::CPU_arch); |
172 | if (Attr.hasValue()) |
173 | isV7 = Attr.getValue() == ARMBuildAttrs::v7; |
174 | |
175 | Attr = Attributes.getAttributeValue(ARMBuildAttrs::CPU_arch_profile); |
176 | if (Attr.hasValue()) { |
177 | switch (Attr.getValue()) { |
178 | case ARMBuildAttrs::ApplicationProfile: |
179 | Features.AddFeature("aclass"); |
180 | break; |
181 | case ARMBuildAttrs::RealTimeProfile: |
182 | Features.AddFeature("rclass"); |
183 | if (isV7) |
184 | Features.AddFeature("hwdiv"); |
185 | break; |
186 | case ARMBuildAttrs::MicroControllerProfile: |
187 | Features.AddFeature("mclass"); |
188 | if (isV7) |
189 | Features.AddFeature("hwdiv"); |
190 | break; |
191 | } |
192 | } |
193 | |
194 | Attr = Attributes.getAttributeValue(ARMBuildAttrs::THUMB_ISA_use); |
195 | if (Attr.hasValue()) { |
196 | switch (Attr.getValue()) { |
197 | default: |
198 | break; |
199 | case ARMBuildAttrs::Not_Allowed: |
200 | Features.AddFeature("thumb", false); |
201 | Features.AddFeature("thumb2", false); |
202 | break; |
203 | case ARMBuildAttrs::AllowThumb32: |
204 | Features.AddFeature("thumb2"); |
205 | break; |
206 | } |
207 | } |
208 | |
209 | Attr = Attributes.getAttributeValue(ARMBuildAttrs::FP_arch); |
210 | if (Attr.hasValue()) { |
211 | switch (Attr.getValue()) { |
212 | default: |
213 | break; |
214 | case ARMBuildAttrs::Not_Allowed: |
215 | Features.AddFeature("vfp2sp", false); |
216 | Features.AddFeature("vfp3d16sp", false); |
217 | Features.AddFeature("vfp4d16sp", false); |
218 | break; |
219 | case ARMBuildAttrs::AllowFPv2: |
220 | Features.AddFeature("vfp2"); |
221 | break; |
222 | case ARMBuildAttrs::AllowFPv3A: |
223 | case ARMBuildAttrs::AllowFPv3B: |
224 | Features.AddFeature("vfp3"); |
225 | break; |
226 | case ARMBuildAttrs::AllowFPv4A: |
227 | case ARMBuildAttrs::AllowFPv4B: |
228 | Features.AddFeature("vfp4"); |
229 | break; |
230 | } |
231 | } |
232 | |
233 | Attr = Attributes.getAttributeValue(ARMBuildAttrs::Advanced_SIMD_arch); |
234 | if (Attr.hasValue()) { |
235 | switch (Attr.getValue()) { |
236 | default: |
237 | break; |
238 | case ARMBuildAttrs::Not_Allowed: |
239 | Features.AddFeature("neon", false); |
240 | Features.AddFeature("fp16", false); |
241 | break; |
242 | case ARMBuildAttrs::AllowNeon: |
243 | Features.AddFeature("neon"); |
244 | break; |
245 | case ARMBuildAttrs::AllowNeon2: |
246 | Features.AddFeature("neon"); |
247 | Features.AddFeature("fp16"); |
248 | break; |
249 | } |
250 | } |
251 | |
252 | Attr = Attributes.getAttributeValue(ARMBuildAttrs::MVE_arch); |
253 | if (Attr.hasValue()) { |
254 | switch (Attr.getValue()) { |
255 | default: |
256 | break; |
257 | case ARMBuildAttrs::Not_Allowed: |
258 | Features.AddFeature("mve", false); |
259 | Features.AddFeature("mve.fp", false); |
260 | break; |
261 | case ARMBuildAttrs::AllowMVEInteger: |
262 | Features.AddFeature("mve.fp", false); |
263 | Features.AddFeature("mve"); |
264 | break; |
265 | case ARMBuildAttrs::AllowMVEIntegerAndFloat: |
266 | Features.AddFeature("mve.fp"); |
267 | break; |
268 | } |
269 | } |
270 | |
271 | Attr = Attributes.getAttributeValue(ARMBuildAttrs::DIV_use); |
272 | if (Attr.hasValue()) { |
273 | switch (Attr.getValue()) { |
274 | default: |
275 | break; |
276 | case ARMBuildAttrs::DisallowDIV: |
277 | Features.AddFeature("hwdiv", false); |
278 | Features.AddFeature("hwdiv-arm", false); |
279 | break; |
280 | case ARMBuildAttrs::AllowDIVExt: |
281 | Features.AddFeature("hwdiv"); |
282 | Features.AddFeature("hwdiv-arm"); |
283 | break; |
284 | } |
285 | } |
286 | |
287 | return Features; |
288 | } |
289 | |
290 | SubtargetFeatures ELFObjectFileBase::getRISCVFeatures() const { |
291 | SubtargetFeatures Features; |
292 | unsigned PlatformFlags = getPlatformFlags(); |
293 | |
294 | if (PlatformFlags & ELF::EF_RISCV_RVC) { |
295 | Features.AddFeature("c"); |
296 | } |
297 | |
298 | |
299 | |
300 | RISCVAttributeParser Attributes; |
301 | if (Error E = getBuildAttributes(Attributes)) { |
302 | |
303 | consumeError(std::move(E)); |
304 | return Features; |
305 | } |
306 | |
307 | Optional<StringRef> Attr = Attributes.getAttributeString(RISCVAttrs::ARCH); |
308 | if (Attr.hasValue()) { |
309 | |
310 | |
311 | |
312 | StringRef Arch = Attr.getValue(); |
313 | if (Arch.consume_front("rv32")) |
314 | Features.AddFeature("64bit", false); |
315 | else if (Arch.consume_front("rv64")) |
316 | Features.AddFeature("64bit"); |
317 | |
318 | while (!Arch.empty()) { |
319 | switch (Arch[0]) { |
320 | default: |
321 | break; |
322 | case 'i': |
323 | Features.AddFeature("e", false); |
324 | break; |
325 | case 'd': |
326 | Features.AddFeature("f"); |
327 | LLVM_FALLTHROUGH; |
328 | case 'e': |
329 | case 'm': |
330 | case 'a': |
331 | case 'f': |
332 | case 'c': |
333 | Features.AddFeature(Arch.take_front()); |
334 | break; |
335 | } |
336 | |
337 | |
338 | Arch = Arch.drop_until([](char c) { return c == '_' || c == '\0'; }); |
339 | Arch = Arch.drop_while([](char c) { return c == '_'; }); |
340 | } |
341 | } |
342 | |
343 | return Features; |
344 | } |
345 | |
346 | SubtargetFeatures ELFObjectFileBase::getFeatures() const { |
347 | switch (getEMachine()) { |
348 | case ELF::EM_MIPS: |
349 | return getMIPSFeatures(); |
350 | case ELF::EM_ARM: |
351 | return getARMFeatures(); |
352 | case ELF::EM_RISCV: |
353 | return getRISCVFeatures(); |
354 | default: |
355 | return SubtargetFeatures(); |
356 | } |
357 | } |
358 | |
359 | Optional<StringRef> ELFObjectFileBase::tryGetCPUName() const { |
360 | switch (getEMachine()) { |
361 | case ELF::EM_AMDGPU: |
362 | return getAMDGPUCPUName(); |
363 | default: |
364 | return None; |
365 | } |
366 | } |
367 | |
368 | StringRef ELFObjectFileBase::getAMDGPUCPUName() const { |
369 | assert(getEMachine() == ELF::EM_AMDGPU); |
370 | unsigned CPU = getPlatformFlags() & ELF::EF_AMDGPU_MACH; |
371 | |
372 | switch (CPU) { |
373 | |
374 | case ELF::EF_AMDGPU_MACH_R600_R600: |
375 | return "r600"; |
376 | case ELF::EF_AMDGPU_MACH_R600_R630: |
377 | return "r630"; |
378 | case ELF::EF_AMDGPU_MACH_R600_RS880: |
379 | return "rs880"; |
380 | case ELF::EF_AMDGPU_MACH_R600_RV670: |
381 | return "rv670"; |
382 | |
383 | |
384 | case ELF::EF_AMDGPU_MACH_R600_RV710: |
385 | return "rv710"; |
386 | case ELF::EF_AMDGPU_MACH_R600_RV730: |
387 | return "rv730"; |
388 | case ELF::EF_AMDGPU_MACH_R600_RV770: |
389 | return "rv770"; |
390 | |
391 | |
392 | case ELF::EF_AMDGPU_MACH_R600_CEDAR: |
393 | return "cedar"; |
394 | case ELF::EF_AMDGPU_MACH_R600_CYPRESS: |
395 | return "cypress"; |
396 | case ELF::EF_AMDGPU_MACH_R600_JUNIPER: |
397 | return "juniper"; |
398 | case ELF::EF_AMDGPU_MACH_R600_REDWOOD: |
399 | return "redwood"; |
400 | case ELF::EF_AMDGPU_MACH_R600_SUMO: |
401 | return "sumo"; |
402 | |
403 | |
404 | case ELF::EF_AMDGPU_MACH_R600_BARTS: |
405 | return "barts"; |
406 | case ELF::EF_AMDGPU_MACH_R600_CAICOS: |
407 | return "caicos"; |
408 | case ELF::EF_AMDGPU_MACH_R600_CAYMAN: |
409 | return "cayman"; |
410 | case ELF::EF_AMDGPU_MACH_R600_TURKS: |
411 | return "turks"; |
412 | |
413 | |
414 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX600: |
415 | return "gfx600"; |
416 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX601: |
417 | return "gfx601"; |
418 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX602: |
419 | return "gfx602"; |
420 | |
421 | |
422 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX700: |
423 | return "gfx700"; |
424 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX701: |
425 | return "gfx701"; |
426 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX702: |
427 | return "gfx702"; |
428 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX703: |
429 | return "gfx703"; |
430 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX704: |
431 | return "gfx704"; |
432 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX705: |
433 | return "gfx705"; |
434 | |
435 | |
436 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX801: |
437 | return "gfx801"; |
438 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX802: |
439 | return "gfx802"; |
440 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX803: |
441 | return "gfx803"; |
442 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX805: |
443 | return "gfx805"; |
444 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX810: |
445 | return "gfx810"; |
446 | |
447 | |
448 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX900: |
449 | return "gfx900"; |
450 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX902: |
451 | return "gfx902"; |
452 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX904: |
453 | return "gfx904"; |
454 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX906: |
455 | return "gfx906"; |
456 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX908: |
457 | return "gfx908"; |
458 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX909: |
459 | return "gfx909"; |
460 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX90A: |
461 | return "gfx90a"; |
462 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX90C: |
463 | return "gfx90c"; |
464 | |
465 | |
466 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1010: |
467 | return "gfx1010"; |
468 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1011: |
469 | return "gfx1011"; |
470 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1012: |
471 | return "gfx1012"; |
472 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1013: |
473 | return "gfx1013"; |
474 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1030: |
475 | return "gfx1030"; |
476 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1031: |
477 | return "gfx1031"; |
478 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1032: |
479 | return "gfx1032"; |
480 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1033: |
481 | return "gfx1033"; |
482 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1034: |
483 | return "gfx1034"; |
484 | case ELF::EF_AMDGPU_MACH_AMDGCN_GFX1035: |
485 | return "gfx1035"; |
486 | default: |
487 | llvm_unreachable("Unknown EF_AMDGPU_MACH value"); |
488 | } |
489 | } |
490 | |
491 | |
492 | void ELFObjectFileBase::setARMSubArch(Triple &TheTriple) const { |
493 | if (TheTriple.getSubArch() != Triple::NoSubArch) |
494 | return; |
495 | |
496 | ARMAttributeParser Attributes; |
497 | if (Error E = getBuildAttributes(Attributes)) { |
498 | |
499 | consumeError(std::move(E)); |
500 | return; |
501 | } |
502 | |
503 | std::string Triple; |
504 | |
505 | if (TheTriple.isThumb()) |
506 | Triple = "thumb"; |
507 | else |
508 | Triple = "arm"; |
509 | |
510 | Optional<unsigned> Attr = |
511 | Attributes.getAttributeValue(ARMBuildAttrs::CPU_arch); |
512 | if (Attr.hasValue()) { |
513 | switch (Attr.getValue()) { |
514 | case ARMBuildAttrs::v4: |
515 | Triple += "v4"; |
516 | break; |
517 | case ARMBuildAttrs::v4T: |
518 | Triple += "v4t"; |
519 | break; |
520 | case ARMBuildAttrs::v5T: |
521 | Triple += "v5t"; |
522 | break; |
523 | case ARMBuildAttrs::v5TE: |
524 | Triple += "v5te"; |
525 | break; |
526 | case ARMBuildAttrs::v5TEJ: |
527 | Triple += "v5tej"; |
528 | break; |
529 | case ARMBuildAttrs::v6: |
530 | Triple += "v6"; |
531 | break; |
532 | case ARMBuildAttrs::v6KZ: |
533 | Triple += "v6kz"; |
534 | break; |
535 | case ARMBuildAttrs::v6T2: |
536 | Triple += "v6t2"; |
537 | break; |
538 | case ARMBuildAttrs::v6K: |
539 | Triple += "v6k"; |
540 | break; |
541 | case ARMBuildAttrs::v7: |
542 | Triple += "v7"; |
543 | break; |
544 | case ARMBuildAttrs::v6_M: |
545 | Triple += "v6m"; |
546 | break; |
547 | case ARMBuildAttrs::v6S_M: |
548 | Triple += "v6sm"; |
549 | break; |
550 | case ARMBuildAttrs::v7E_M: |
551 | Triple += "v7em"; |
552 | break; |
553 | case ARMBuildAttrs::v8_A: |
554 | Triple += "v8a"; |
555 | break; |
556 | case ARMBuildAttrs::v8_R: |
557 | Triple += "v8r"; |
558 | break; |
559 | case ARMBuildAttrs::v8_M_Base: |
560 | Triple += "v8m.base"; |
561 | break; |
562 | case ARMBuildAttrs::v8_M_Main: |
563 | Triple += "v8m.main"; |
564 | break; |
565 | case ARMBuildAttrs::v8_1_M_Main: |
566 | Triple += "v8.1m.main"; |
567 | break; |
568 | } |
569 | } |
570 | if (!isLittleEndian()) |
571 | Triple += "eb"; |
572 | |
573 | TheTriple.setArchName(Triple); |
574 | } |
575 | |
576 | std::vector<std::pair<Optional<DataRefImpl>, uint64_t>> |
577 | ELFObjectFileBase::getPltAddresses() const { |
578 | std::string Err; |
579 | const auto Triple = makeTriple(); |
580 | const auto *T = TargetRegistry::lookupTarget(Triple.str(), Err); |
581 | if (!T) |
582 | return {}; |
583 | uint64_t JumpSlotReloc = 0; |
584 | switch (Triple.getArch()) { |
585 | case Triple::x86: |
586 | JumpSlotReloc = ELF::R_386_JUMP_SLOT; |
587 | break; |
588 | case Triple::x86_64: |
589 | JumpSlotReloc = ELF::R_X86_64_JUMP_SLOT; |
590 | break; |
591 | case Triple::aarch64: |
592 | case Triple::aarch64_be: |
593 | JumpSlotReloc = ELF::R_AARCH64_JUMP_SLOT; |
594 | break; |
595 | default: |
596 | return {}; |
597 | } |
598 | std::unique_ptr<const MCInstrInfo> MII(T->createMCInstrInfo()); |
599 | std::unique_ptr<const MCInstrAnalysis> MIA( |
600 | T->createMCInstrAnalysis(MII.get())); |
601 | if (!MIA) |
602 | return {}; |
603 | Optional<SectionRef> Plt = None, RelaPlt = None, GotPlt = None; |
604 | for (const SectionRef &Section : sections()) { |
605 | Expected<StringRef> NameOrErr = Section.getName(); |
606 | if (!NameOrErr) { |
607 | consumeError(NameOrErr.takeError()); |
608 | continue; |
609 | } |
610 | StringRef Name = *NameOrErr; |
611 | |
612 | if (Name == ".plt") |
613 | Plt = Section; |
614 | else if (Name == ".rela.plt" || Name == ".rel.plt") |
615 | RelaPlt = Section; |
616 | else if (Name == ".got.plt") |
617 | GotPlt = Section; |
618 | } |
619 | if (!Plt || !RelaPlt || !GotPlt) |
620 | return {}; |
621 | Expected<StringRef> PltContents = Plt->getContents(); |
622 | if (!PltContents) { |
623 | consumeError(PltContents.takeError()); |
624 | return {}; |
625 | } |
626 | auto PltEntries = MIA->findPltEntries(Plt->getAddress(), |
627 | arrayRefFromStringRef(*PltContents), |
628 | GotPlt->getAddress(), Triple); |
629 | |
630 | DenseMap<uint64_t, uint64_t> GotToPlt; |
631 | for (const auto &Entry : PltEntries) |
632 | GotToPlt.insert(std::make_pair(Entry.second, Entry.first)); |
633 | |
634 | |
635 | std::vector<std::pair<Optional<DataRefImpl>, uint64_t>> Result; |
636 | for (const auto &Relocation : RelaPlt->relocations()) { |
637 | if (Relocation.getType() != JumpSlotReloc) |
638 | continue; |
639 | auto PltEntryIter = GotToPlt.find(Relocation.getOffset()); |
640 | if (PltEntryIter != GotToPlt.end()) { |
641 | symbol_iterator Sym = Relocation.getSymbol(); |
642 | if (Sym == symbol_end()) |
643 | Result.emplace_back(None, PltEntryIter->second); |
644 | else |
645 | Result.emplace_back(Sym->getRawDataRefImpl(), PltEntryIter->second); |
646 | } |
647 | } |
648 | return Result; |
649 | } |
1 | |
2 | |
3 | |
4 | |
5 | |
6 | |
7 | |
8 | |
9 | |
10 | |
11 | |
12 | |
13 | #ifndef LLVM_SUPPORT_MATHEXTRAS_H |
14 | #define LLVM_SUPPORT_MATHEXTRAS_H |
15 | |
16 | #include "llvm/Support/Compiler.h" |
17 | #include <cassert> |
18 | #include <climits> |
19 | #include <cmath> |
20 | #include <cstdint> |
21 | #include <cstring> |
22 | #include <limits> |
23 | #include <type_traits> |
24 | |
25 | #ifdef __ANDROID_NDK__ |
26 | #include <android/api-level.h> |
27 | #endif |
28 | |
29 | #ifdef _MSC_VER |
30 | |
31 | |
32 | |
33 | extern "C" { |
34 | unsigned char _BitScanForward(unsigned long *_Index, unsigned long _Mask); |
35 | unsigned char _BitScanForward64(unsigned long *_Index, unsigned __int64 _Mask); |
36 | unsigned char _BitScanReverse(unsigned long *_Index, unsigned long _Mask); |
37 | unsigned char _BitScanReverse64(unsigned long *_Index, unsigned __int64 _Mask); |
38 | } |
39 | #endif |
40 | |
41 | namespace llvm { |
42 | |
43 | |
44 | enum ZeroBehavior { |
45 | |
46 | ZB_Undefined, |
47 | |
48 | ZB_Max, |
49 | |
50 | ZB_Width |
51 | }; |
52 | |
53 | |
54 | namespace numbers { |
55 | |
56 | |
57 | constexpr double e = 2.7182818284590452354, |
58 | egamma = .57721566490153286061, |
59 | ln2 = .69314718055994530942, |
60 | ln10 = 2.3025850929940456840, |
61 | log2e = 1.4426950408889634074, |
62 | log10e = .43429448190325182765, |
63 | pi = 3.1415926535897932385, |
64 | inv_pi = .31830988618379067154, |
65 | sqrtpi = 1.7724538509055160273, |
66 | inv_sqrtpi = .56418958354775628695, |
67 | sqrt2 = 1.4142135623730950488, |
68 | inv_sqrt2 = .70710678118654752440, |
69 | sqrt3 = 1.7320508075688772935, |
70 | inv_sqrt3 = .57735026918962576451, |
71 | phi = 1.6180339887498948482; |
72 | constexpr float ef = 2.71828183F, |
73 | egammaf = .577215665F, |
74 | ln2f = .693147181F, |
75 | ln10f = 2.30258509F, |
76 | log2ef = 1.44269504F, |
77 | log10ef = .434294482F, |
78 | pif = 3.14159265F, |
79 | inv_pif = .318309886F, |
80 | sqrtpif = 1.77245385F, |
81 | inv_sqrtpif = .564189584F, |
82 | sqrt2f = 1.41421356F, |
83 | inv_sqrt2f = .707106781F, |
84 | sqrt3f = 1.73205081F, |
85 | inv_sqrt3f = .577350269F, |
86 | phif = 1.61803399F; |
87 | } |
88 | |
89 | namespace detail { |
90 | template <typename T, std::size_t SizeOfT> struct TrailingZerosCounter { |
91 | static unsigned count(T Val, ZeroBehavior) { |
92 | if (!Val) |
93 | return std::numeric_limits<T>::digits; |
94 | if (Val & 0x1) |
95 | return 0; |
96 | |
97 | |
98 | unsigned ZeroBits = 0; |
99 | T Shift = std::numeric_limits<T>::digits >> 1; |
100 | T Mask = std::numeric_limits<T>::max() >> Shift; |
101 | while (Shift) { |
102 | if ((Val & Mask) == 0) { |
103 | Val >>= Shift; |
104 | ZeroBits |= Shift; |
105 | } |
106 | Shift >>= 1; |
107 | Mask >>= Shift; |
108 | } |
109 | return ZeroBits; |
110 | } |
111 | }; |
112 | |
113 | #if defined(__GNUC__) || defined(_MSC_VER) |
114 | template <typename T> struct TrailingZerosCounter<T, 4> { |
115 | static unsigned count(T Val, ZeroBehavior ZB) { |
116 | if (ZB != ZB_Undefined && Val == 0) |
117 | return 32; |
118 | |
119 | #if __has_builtin(__builtin_ctz) || defined(__GNUC__) |
120 | return __builtin_ctz(Val); |
121 | #elif defined(_MSC_VER) |
122 | unsigned long Index; |
123 | _BitScanForward(&Index, Val); |
124 | return Index; |
125 | #endif |
126 | } |
127 | }; |
128 | |
129 | #if !defined(_MSC_VER) || defined(_M_X64) |
130 | template <typename T> struct TrailingZerosCounter<T, 8> { |
131 | static unsigned count(T Val, ZeroBehavior ZB) { |
132 | if (ZB != ZB_Undefined && Val == 0) |
| 3 | | Assuming 'Val' is equal to 0 | |
|
| |
133 | return 64; |
| |
134 | |
135 | #if __has_builtin(__builtin_ctzll) || defined(__GNUC__) |
136 | return __builtin_ctzll(Val); |
137 | #elif defined(_MSC_VER) |
138 | unsigned long Index; |
139 | _BitScanForward64(&Index, Val); |
140 | return Index; |
141 | #endif |
142 | } |
143 | }; |
144 | #endif |
145 | #endif |
146 | } |
147 | |
148 | |
149 | |
150 | |
151 | |
152 | |
153 | |
154 | |
155 | template <typename T> |
156 | unsigned countTrailingZeros(T Val, ZeroBehavior ZB = ZB_Width) { |
157 | static_assert(std::numeric_limits<T>::is_integer && |
158 | !std::numeric_limits<T>::is_signed, |
159 | "Only unsigned integral types are allowed."); |
160 | return llvm::detail::TrailingZerosCounter<T, sizeof(T)>::count(Val, ZB); |
| 2 | | Calling 'TrailingZerosCounter::count' | |
|
| 6 | | Returning from 'TrailingZerosCounter::count' | |
|
| |
161 | } |
162 | |
163 | namespace detail { |
164 | template <typename T, std::size_t SizeOfT> struct LeadingZerosCounter { |
165 | static unsigned count(T Val, ZeroBehavior) { |
166 | if (!Val) |
167 | return std::numeric_limits<T>::digits; |
168 | |
169 | |
170 | unsigned ZeroBits = 0; |
171 | for (T Shift = std::numeric_limits<T>::digits >> 1; Shift; Shift >>= 1) { |
172 | T Tmp = Val >> Shift; |
173 | if (Tmp) |
174 | Val = Tmp; |
175 | else |
176 | ZeroBits |= Shift; |
177 | } |
178 | return ZeroBits; |
179 | } |
180 | }; |
181 | |
182 | #if defined(__GNUC__) || defined(_MSC_VER) |
183 | template <typename T> struct LeadingZerosCounter<T, 4> { |
184 | static unsigned count(T Val, ZeroBehavior ZB) { |
185 | if (ZB != ZB_Undefined && Val == 0) |
186 | return 32; |
187 | |
188 | #if __has_builtin(__builtin_clz) || defined(__GNUC__) |
189 | return __builtin_clz(Val); |
190 | #elif defined(_MSC_VER) |
191 | unsigned long Index; |
192 | _BitScanReverse(&Index, Val); |
193 | return Index ^ 31; |
194 | #endif |
195 | } |
196 | }; |
197 | |
198 | #if !defined(_MSC_VER) || defined(_M_X64) |
199 | template <typename T> struct LeadingZerosCounter<T, 8> { |
200 | static unsigned count(T Val, ZeroBehavior ZB) { |
201 | if (ZB != ZB_Undefined && Val == 0) |
202 | return 64; |
203 | |
204 | #if __has_builtin(__builtin_clzll) || defined(__GNUC__) |
205 | return __builtin_clzll(Val); |
206 | #elif defined(_MSC_VER) |
207 | unsigned long Index; |
208 | _BitScanReverse64(&Index, Val); |
209 | return Index ^ 63; |
210 | #endif |
211 | } |
212 | }; |
213 | #endif |
214 | #endif |
215 | } |
216 | |
217 | |
218 | |
219 | |
220 | |
221 | |
222 | |
223 | |
224 | template <typename T> |
225 | unsigned countLeadingZeros(T Val, ZeroBehavior ZB = ZB_Width) { |
226 | static_assert(std::numeric_limits<T>::is_integer && |
227 | !std::numeric_limits<T>::is_signed, |
228 | "Only unsigned integral types are allowed."); |
229 | return llvm::detail::LeadingZerosCounter<T, sizeof(T)>::count(Val, ZB); |
230 | } |
231 | |
232 | |
233 | |
234 | |
235 | |
236 | |
237 | |
238 | |
239 | template <typename T> T findFirstSet(T Val, ZeroBehavior ZB = ZB_Max) { |
240 | if (ZB == ZB_Max && Val == 0) |
241 | return std::numeric_limits<T>::max(); |
242 | |
243 | return countTrailingZeros(Val, ZB_Undefined); |
244 | } |
245 | |
246 | |
247 | |
248 | template <typename T> T maskTrailingOnes(unsigned N) { |
249 | static_assert(std::is_unsigned<T>::value, "Invalid type!"); |
250 | const unsigned Bits = CHAR_BIT * sizeof(T); |
251 | assert(N <= Bits && "Invalid bit index"); |
252 | return N == 0 ? 0 : (T(-1) >> (Bits - N)); |
253 | } |
254 | |
255 | |
256 | |
257 | template <typename T> T maskLeadingOnes(unsigned N) { |
258 | return ~maskTrailingOnes<T>(CHAR_BIT * sizeof(T) - N); |
259 | } |
260 | |
261 | |
262 | |
263 | template <typename T> T maskTrailingZeros(unsigned N) { |
264 | return maskLeadingOnes<T>(CHAR_BIT * sizeof(T) - N); |
265 | } |
266 | |
267 | |
268 | |
269 | template <typename T> T maskLeadingZeros(unsigned N) { |
270 | return maskTrailingOnes<T>(CHAR_BIT * sizeof(T) - N); |
271 | } |
272 | |
273 | |
274 | |
275 | |
276 | |
277 | |
278 | |
279 | |
280 | template <typename T> T findLastSet(T Val, ZeroBehavior ZB = ZB_Max) { |
281 | if (ZB == ZB_Max && Val == 0) |
282 | return std::numeric_limits<T>::max(); |
283 | |
284 | |
285 | |
286 | return countLeadingZeros(Val, ZB_Undefined) ^ |
287 | (std::numeric_limits<T>::digits - 1); |
288 | } |
289 | |
290 | |
291 | |
292 | |
293 | static const unsigned char BitReverseTable256[256] = { |
294 | #define R2(n) n, n + 2 * 64, n + 1 * 64, n + 3 * 64 |
295 | #define R4(n) R2(n), R2(n + 2 * 16), R2(n + 1 * 16), R2(n + 3 * 16) |
296 | #define R6(n) R4(n), R4(n + 2 * 4), R4(n + 1 * 4), R4(n + 3 * 4) |
297 | R6(0), R6(2), R6(1), R6(3) |
298 | #undef R2 |
299 | #undef R4 |
300 | #undef R6 |
301 | }; |
302 | |
303 | |
304 | template <typename T> |
305 | T reverseBits(T Val) { |
306 | unsigned char in[sizeof(Val)]; |
307 | unsigned char out[sizeof(Val)]; |
308 | std::memcpy(in, &Val, sizeof(Val)); |
309 | for (unsigned i = 0; i < sizeof(Val); ++i) |
310 | out[(sizeof(Val) - i) - 1] = BitReverseTable256[in[i]]; |
311 | std::memcpy(&Val, out, sizeof(Val)); |
312 | return Val; |
313 | } |
314 | |
315 | #if __has_builtin(__builtin_bitreverse8) |
316 | template<> |
317 | inline uint8_t reverseBits<uint8_t>(uint8_t Val) { |
318 | return __builtin_bitreverse8(Val); |
319 | } |
320 | #endif |
321 | |
322 | #if __has_builtin(__builtin_bitreverse16) |
323 | template<> |
324 | inline uint16_t reverseBits<uint16_t>(uint16_t Val) { |
325 | return __builtin_bitreverse16(Val); |
326 | } |
327 | #endif |
328 | |
329 | #if __has_builtin(__builtin_bitreverse32) |
330 | template<> |
331 | inline uint32_t reverseBits<uint32_t>(uint32_t Val) { |
332 | return __builtin_bitreverse32(Val); |
333 | } |
334 | #endif |
335 | |
336 | #if __has_builtin(__builtin_bitreverse64) |
337 | template<> |
338 | inline uint64_t reverseBits<uint64_t>(uint64_t Val) { |
339 | return __builtin_bitreverse64(Val); |
340 | } |
341 | #endif |
342 | |
343 | |
344 | |
345 | |
346 | |
347 | |
348 | constexpr inline uint32_t Hi_32(uint64_t Value) { |
349 | return static_cast<uint32_t>(Value >> 32); |
350 | } |
351 | |
352 | |
353 | constexpr inline uint32_t Lo_32(uint64_t Value) { |
354 | return static_cast<uint32_t>(Value); |
355 | } |
356 | |
357 | |
358 | constexpr inline uint64_t Make_64(uint32_t High, uint32_t Low) { |
359 | return ((uint64_t)High << 32) | (uint64_t)Low; |
360 | } |
361 | |
362 | |
363 | template <unsigned N> constexpr inline bool isInt(int64_t x) { |
364 | return N >= 64 || (-(INT64_C(1)<<(N-1)) <= x && x < (INT64_C(1)<<(N-1))); |
365 | } |
366 | |
367 | template <> constexpr inline bool isInt<8>(int64_t x) { |
368 | return static_cast<int8_t>(x) == x; |
369 | } |
370 | template <> constexpr inline bool isInt<16>(int64_t x) { |
371 | return static_cast<int16_t>(x) == x; |
372 | } |
373 | template <> constexpr inline bool isInt<32>(int64_t x) { |
374 | return static_cast<int32_t>(x) == x; |
375 | } |
376 | |
377 | |
378 | template <unsigned N, unsigned S> |
379 | constexpr inline bool isShiftedInt(int64_t x) { |
380 | static_assert( |
381 | N > 0, "isShiftedInt<0> doesn't make sense (refers to a 0-bit number."); |
382 | static_assert(N + S <= 64, "isShiftedInt<N, S> with N + S > 64 is too wide."); |
383 | return isInt<N + S>(x) && (x % (UINT64_C(1) << S) == 0); |
384 | } |
385 | |
386 | |
387 | |
388 | |
389 | |
390 | |
391 | |
392 | |
393 | |
394 | template <unsigned N> |
395 | constexpr inline std::enable_if_t<(N < 64), bool> isUInt(uint64_t X) { |
396 | static_assert(N > 0, "isUInt<0> doesn't make sense"); |
397 | return X < (UINT64_C(1) << (N)); |
398 | } |
399 | template <unsigned N> |
400 | constexpr inline std::enable_if_t<N >= 64, bool> isUInt(uint64_t) { |
401 | return true; |
402 | } |
403 | |
404 | |
405 | template <> constexpr inline bool isUInt<8>(uint64_t x) { |
406 | return static_cast<uint8_t>(x) == x; |
407 | } |
408 | template <> constexpr inline bool isUInt<16>(uint64_t x) { |
409 | return static_cast<uint16_t>(x) == x; |
410 | } |
411 | template <> constexpr inline bool isUInt<32>(uint64_t x) { |
412 | return static_cast<uint32_t>(x) == x; |
413 | } |
414 | |
415 | |
416 | template <unsigned N, unsigned S> |
417 | constexpr inline bool isShiftedUInt(uint64_t x) { |
418 | static_assert( |
419 | N > 0, "isShiftedUInt<0> doesn't make sense (refers to a 0-bit number)"); |
420 | static_assert(N + S <= 64, |
421 | "isShiftedUInt<N, S> with N + S > 64 is too wide."); |
422 | |
423 | |
424 | return isUInt<N + S>(x) && (x % (UINT64_C(1) << S) == 0); |
425 | } |
426 | |
427 | |
428 | inline uint64_t maxUIntN(uint64_t N) { |
429 | assert(N > 0 && N <= 64 && "integer width out of range"); |
430 | |
431 | |
432 | |
433 | |
434 | |
435 | return UINT64_MAX >> (64 - N); |
436 | } |
437 | |
438 | |
439 | inline int64_t minIntN(int64_t N) { |
440 | assert(N > 0 && N <= 64 && "integer width out of range"); |
441 | |
442 | return UINT64_C(1) + ~(UINT64_C(1) << (N - 1)); |
443 | } |
444 | |
445 | |
446 | inline int64_t maxIntN(int64_t N) { |
447 | assert(N > 0 && N <= 64 && "integer width out of range"); |
448 | |
449 | |
450 | |
451 | return (UINT64_C(1) << (N - 1)) - 1; |
452 | } |
453 | |
454 | |
455 | inline bool isUIntN(unsigned N, uint64_t x) { |
456 | return N >= 64 || x <= maxUIntN(N); |
457 | } |
458 | |
459 | |
460 | inline bool isIntN(unsigned N, int64_t x) { |
461 | return N >= 64 || (minIntN(N) <= x && x <= maxIntN(N)); |
462 | } |
463 | |
464 | |
465 | |
466 | |
467 | constexpr inline bool isMask_32(uint32_t Value) { |
468 | return Value && ((Value + 1) & Value) == 0; |
469 | } |
470 | |
471 | |
472 | |
473 | constexpr inline bool isMask_64(uint64_t Value) { |
474 | return Value && ((Value + 1) & Value) == 0; |
475 | } |
476 | |
477 | |
478 | |
479 | constexpr inline bool isShiftedMask_32(uint32_t Value) { |
480 | return Value && isMask_32((Value - 1) | Value); |
481 | } |
482 | |
483 | |
484 | |
485 | constexpr inline bool isShiftedMask_64(uint64_t Value) { |
486 | return Value && isMask_64((Value - 1) | Value); |
487 | } |
488 | |
489 | |
490 | |
491 | constexpr inline bool isPowerOf2_32(uint32_t Value) { |
492 | return Value && !(Value & (Value - 1)); |
493 | } |
494 | |
495 | |
496 | constexpr inline bool isPowerOf2_64(uint64_t Value) { |
497 | return Value && !(Value & (Value - 1)); |
498 | } |
499 | |
500 | |
501 | |
502 | |
503 | |
504 | |
505 | |
506 | |
507 | |
508 | template <typename T> |
509 | unsigned countLeadingOnes(T Value, ZeroBehavior ZB = ZB_Width) { |
510 | static_assert(std::numeric_limits<T>::is_integer && |
511 | !std::numeric_limits<T>::is_signed, |
512 | "Only unsigned integral types are allowed."); |
513 | return countLeadingZeros<T>(~Value, ZB); |
514 | } |
515 | |
516 | |
517 | |
518 | |
519 | |
520 | |
521 | |
522 | |
523 | |
524 | template <typename T> |
525 | unsigned countTrailingOnes(T Value, ZeroBehavior ZB = ZB_Width) { |
526 | static_assert(std::numeric_limits<T>::is_integer && |
527 | !std::numeric_limits<T>::is_signed, |
528 | "Only unsigned integral types are allowed."); |
529 | return countTrailingZeros<T>(~Value, ZB); |
530 | } |
531 | |
532 | namespace detail { |
533 | template <typename T, std::size_t SizeOfT> struct PopulationCounter { |
534 | static unsigned count(T Value) { |
535 | |
536 | static_assert(SizeOfT <= 4, "Not implemented!"); |
537 | #if defined(__GNUC__) |
538 | return __builtin_popcount(Value); |
539 | #else |
540 | uint32_t v = Value; |
541 | v = v - ((v >> 1) & 0x55555555); |
542 | v = (v & 0x33333333) + ((v >> 2) & 0x33333333); |
543 | return ((v + (v >> 4) & 0xF0F0F0F) * 0x1010101) >> 24; |
544 | #endif |
545 | } |
546 | }; |
547 | |
548 | template <typename T> struct PopulationCounter<T, 8> { |
549 | static unsigned count(T Value) { |
550 | #if defined(__GNUC__) |
551 | return __builtin_popcountll(Value); |
552 | #else |
553 | uint64_t v = Value; |
554 | v = v - ((v >> 1) & 0x5555555555555555ULL); |
555 | v = (v & 0x3333333333333333ULL) + ((v >> 2) & 0x3333333333333333ULL); |
556 | v = (v + (v >> 4)) & 0x0F0F0F0F0F0F0F0FULL; |
557 | return unsigned((uint64_t)(v * 0x0101010101010101ULL) >> 56); |
558 | #endif |
559 | } |
560 | }; |
561 | } |
562 | |
563 | |
564 | |
565 | |
566 | template <typename T> |
567 | inline unsigned countPopulation(T Value) { |
568 | static_assert(std::numeric_limits<T>::is_integer && |
569 | !std::numeric_limits<T>::is_signed, |
570 | "Only unsigned integral types are allowed."); |
571 | return detail::PopulationCounter<T, sizeof(T)>::count(Value); |
572 | } |
573 | |
574 | |
575 | |
576 | template <size_t kValue> constexpr inline size_t CTLog2() { |
577 | static_assert(kValue > 0 && llvm::isPowerOf2_64(kValue), |
578 | "Value is not a valid power of 2"); |
579 | return 1 + CTLog2<kValue / 2>(); |
580 | } |
581 | |
582 | template <> constexpr inline size_t CTLog2<1>() { return 0; } |
583 | |
584 | |
585 | inline double Log2(double Value) { |
586 | #if defined(__ANDROID_API__) && __ANDROID_API__ < 18 |
587 | return __builtin_log(Value) / __builtin_log(2.0); |
588 | #else |
589 | return log2(Value); |
590 | #endif |
591 | } |
592 | |
593 | |
594 | |
595 | |
596 | inline unsigned Log2_32(uint32_t Value) { |
597 | return 31 - countLeadingZeros(Value); |
598 | } |
599 | |
600 | |
601 | |
602 | inline unsigned Log2_64(uint64_t Value) { |
603 | return 63 - countLeadingZeros(Value); |
604 | } |
605 | |
606 | |
607 | |
608 | |
609 | inline unsigned Log2_32_Ceil(uint32_t Value) { |
610 | return 32 - countLeadingZeros(Value - 1); |
611 | } |
612 | |
613 | |
614 | |
615 | inline unsigned Log2_64_Ceil(uint64_t Value) { |
616 | return 64 - countLeadingZeros(Value - 1); |
617 | } |
618 | |
619 | |
620 | template <typename T> |
621 | inline T greatestCommonDivisor(T A, T B) { |
622 | while (B) { |
623 | T Tmp = B; |
624 | B = A % B; |
625 | A = Tmp; |
626 | } |
627 | return A; |
628 | } |
629 | |
630 | inline uint64_t GreatestCommonDivisor64(uint64_t A, uint64_t B) { |
631 | return greatestCommonDivisor<uint64_t>(A, B); |
632 | } |
633 | |
634 | |
635 | inline double BitsToDouble(uint64_t Bits) { |
636 | double D; |
637 | static_assert(sizeof(uint64_t) == sizeof(double), "Unexpected type sizes"); |
638 | memcpy(&D, &Bits, sizeof(Bits)); |
639 | return D; |
640 | } |
641 | |
642 | |
643 | inline float BitsToFloat(uint32_t Bits) { |
644 | float F; |
645 | static_assert(sizeof(uint32_t) == sizeof(float), "Unexpected type sizes"); |
646 | memcpy(&F, &Bits, sizeof(Bits)); |
647 | return F; |
648 | } |
649 | |
650 | |
651 | |
652 | |
653 | inline uint64_t DoubleToBits(double Double) { |
654 | uint64_t Bits; |
655 | static_assert(sizeof(uint64_t) == sizeof(double), "Unexpected type sizes"); |
656 | memcpy(&Bits, &Double, sizeof(Double)); |
657 | return Bits; |
658 | } |
659 | |
660 | |
661 | |
662 | |
663 | inline uint32_t FloatToBits(float Float) { |
664 | uint32_t Bits; |
665 | static_assert(sizeof(uint32_t) == sizeof(float), "Unexpected type sizes"); |
666 | memcpy(&Bits, &Float, sizeof(Float)); |
667 | return Bits; |
668 | } |
669 | |
670 | |
671 | |
672 | constexpr inline uint64_t MinAlign(uint64_t A, uint64_t B) { |
673 | |
674 | |
675 | |
676 | |
677 | |
678 | return (A | B) & (1 + ~(A | B)); |
679 | } |
680 | |
681 | |
682 | |
683 | inline uint64_t NextPowerOf2(uint64_t A) { |
684 | A |= (A >> 1); |
685 | A |= (A >> 2); |
686 | A |= (A >> 4); |
687 | A |= (A >> 8); |
688 | A |= (A >> 16); |
689 | A |= (A >> 32); |
690 | return A + 1; |
691 | } |
692 | |
693 | |
694 | |
695 | inline uint64_t PowerOf2Floor(uint64_t A) { |
696 | if (!A) return 0; |
697 | return 1ull << (63 - countLeadingZeros(A, ZB_Undefined)); |
698 | } |
699 | |
700 | |
701 | |
702 | inline uint64_t PowerOf2Ceil(uint64_t A) { |
703 | if (!A) |
704 | return 0; |
705 | return NextPowerOf2(A - 1); |
706 | } |
707 | |
708 | |
709 | |
710 | |
711 | |
712 | |
713 | |
714 | |
715 | |
716 | |
717 | |
718 | |
719 | |
720 | |
721 | |
722 | |
723 | |
724 | |
725 | |
726 | |
727 | |
728 | inline uint64_t alignTo(uint64_t Value, uint64_t Align, uint64_t Skew = 0) { |
729 | assert(Align != 0u && "Align can't be 0."); |
730 | Skew %= Align; |
731 | return (Value + Align - 1 - Skew) / Align * Align + Skew; |
732 | } |
733 | |
734 | |
735 | |
736 | template <uint64_t Align> constexpr inline uint64_t alignTo(uint64_t Value) { |
737 | static_assert(Align != 0u, "Align must be non-zero"); |
738 | return (Value + Align - 1) / Align * Align; |
739 | } |
740 | |
741 | |
742 | inline uint64_t divideCeil(uint64_t Numerator, uint64_t Denominator) { |
743 | return alignTo(Numerator, Denominator) / Denominator; |
744 | } |
745 | |
746 | |
747 | inline uint64_t divideNearest(uint64_t Numerator, uint64_t Denominator) { |
748 | return (Numerator + (Denominator / 2)) / Denominator; |
749 | } |
750 | |
751 | |
752 | |
753 | inline uint64_t alignDown(uint64_t Value, uint64_t Align, uint64_t Skew = 0) { |
754 | assert(Align != 0u && "Align can't be 0."); |
755 | Skew %= Align; |
756 | return (Value - Skew) / Align * Align + Skew; |
757 | } |
758 | |
759 | |
760 | |
761 | template <unsigned B> constexpr inline int32_t SignExtend32(uint32_t X) { |
762 | static_assert(B > 0, "Bit width can't be 0."); |
763 | static_assert(B <= 32, "Bit width out of range."); |
764 | return int32_t(X << (32 - B)) >> (32 - B); |
765 | } |
766 | |
767 | |
768 | |
769 | inline int32_t SignExtend32(uint32_t X, unsigned B) { |
770 | assert(B > 0 && "Bit width can't be 0."); |
771 | assert(B <= 32 && "Bit width out of range."); |
772 | return int32_t(X << (32 - B)) >> (32 - B); |
773 | } |
774 | |
775 | |
776 | |
777 | template <unsigned B> constexpr inline int64_t SignExtend64(uint64_t x) { |
778 | static_assert(B > 0, "Bit width can't be 0."); |
779 | static_assert(B <= 64, "Bit width out of range."); |
780 | return int64_t(x << (64 - B)) >> (64 - B); |
781 | } |
782 | |
783 | |
784 | |
785 | inline int64_t SignExtend64(uint64_t X, unsigned B) { |
786 | assert(B > 0 && "Bit width can't be 0."); |
787 | assert(B <= 64 && "Bit width out of range."); |
788 | return int64_t(X << (64 - B)) >> (64 - B); |
789 | } |
790 | |
791 | |
792 | |
793 | template <typename T> |
794 | std::enable_if_t<std::is_unsigned<T>::value, T> AbsoluteDifference(T X, T Y) { |
795 | return X > Y ? (X - Y) : (Y - X); |
796 | } |
797 | |
798 | |
799 | |
800 | |
801 | template <typename T> |
802 | std::enable_if_t<std::is_unsigned<T>::value, T> |
803 | SaturatingAdd(T X, T Y, bool *ResultOverflowed = nullptr) { |
804 | bool Dummy; |
805 | bool &Overflowed = ResultOverflowed ? *ResultOverflowed : Dummy; |
806 | |
807 | T Z = X + Y; |
808 | Overflowed = (Z < X || Z < Y); |
809 | if (Overflowed) |
810 | return std::numeric_limits<T>::max(); |
811 | else |
812 | return Z; |
813 | } |
814 | |
815 | |
816 | |
817 | |
818 | template <typename T> |
819 | std::enable_if_t<std::is_unsigned<T>::value, T> |
820 | SaturatingMultiply(T X, T Y, bool *ResultOverflowed = nullptr) { |
821 | bool Dummy; |
822 | bool &Overflowed = ResultOverflowed ? *ResultOverflowed : Dummy; |
823 | |
824 | |
825 | |
826 | |
827 | |
828 | |
829 | Overflowed = false; |
830 | |
831 | |
832 | |
833 | |
834 | int Log2Z = Log2_64(X) + Log2_64(Y); |
835 | const T Max = std::numeric_limits<T>::max(); |
836 | int Log2Max = Log2_64(Max); |
837 | if (Log2Z < Log2Max) { |
838 | return X * Y; |
839 | } |
840 | if (Log2Z > Log2Max) { |
841 | Overflowed = true; |
842 | return Max; |
843 | } |
844 | |
845 | |
846 | |
847 | |
848 | T Z = (X >> 1) * Y; |
849 | if (Z & ~(Max >> 1)) { |
850 | Overflowed = true; |
851 | return Max; |
852 | } |
853 | Z <<= 1; |
854 | if (X & 1) |
855 | return SaturatingAdd(Z, Y, ResultOverflowed); |
856 | |
857 | return Z; |
858 | } |
859 | |
860 | |
861 | |
862 | |
863 | |
864 | template <typename T> |
865 | std::enable_if_t<std::is_unsigned<T>::value, T> |
866 | SaturatingMultiplyAdd(T X, T Y, T A, bool *ResultOverflowed = nullptr) { |
867 | bool Dummy; |
868 | bool &Overflowed = ResultOverflowed ? *ResultOverflowed : Dummy; |
869 | |
870 | T Product = SaturatingMultiply(X, Y, &Overflowed); |
871 | if (Overflowed) |
872 | return Product; |
873 | |
874 | return SaturatingAdd(A, Product, &Overflowed); |
875 | } |
876 | |
877 | |
878 | extern const float huge_valf; |
879 | |
880 | |
881 | |
882 | |
883 | template <typename T> |
884 | std::enable_if_t<std::is_signed<T>::value, T> AddOverflow(T X, T Y, T &Result) { |
885 | #if __has_builtin(__builtin_add_overflow) |
886 | return __builtin_add_overflow(X, Y, &Result); |
887 | #else |
888 | |
889 | using U = std::make_unsigned_t<T>; |
890 | const U UX = static_cast<U>(X); |
891 | const U UY = static_cast<U>(Y); |
892 | const U UResult = UX + UY; |
893 | |
894 | |
895 | Result = static_cast<T>(UResult); |
896 | |
897 | |
898 | if (X > 0 && Y > 0) |
899 | return Result <= 0; |
900 | |
901 | if (X < 0 && Y < 0) |
902 | return Result >= 0; |
903 | return false; |
904 | #endif |
905 | } |
906 | |
907 | |
908 | |
909 | template <typename T> |
910 | std::enable_if_t<std::is_signed<T>::value, T> SubOverflow(T X, T Y, T &Result) { |
911 | #if __has_builtin(__builtin_sub_overflow) |
912 | return __builtin_sub_overflow(X, Y, &Result); |
913 | #else |
914 | |
915 | using U = std::make_unsigned_t<T>; |
916 | const U UX = static_cast<U>(X); |
917 | const U UY = static_cast<U>(Y); |
918 | const U UResult = UX - UY; |
919 | |
920 | |
921 | Result = static_cast<T>(UResult); |
922 | |
923 | |
924 | if (X <= 0 && Y > 0) |
925 | return Result >= 0; |
926 | |
927 | if (X >= 0 && Y < 0) |
928 | return Result <= 0; |
929 | return false; |
930 | #endif |
931 | } |
932 | |
933 | |
934 | |
935 | template <typename T> |
936 | std::enable_if_t<std::is_signed<T>::value, T> MulOverflow(T X, T Y, T &Result) { |
937 | |
938 | using U = std::make_unsigned_t<T>; |
939 | const U UX = X < 0 ? (0 - static_cast<U>(X)) : static_cast<U>(X); |
940 | const U UY = Y < 0 ? (0 - static_cast<U>(Y)) : static_cast<U>(Y); |
941 | const U UResult = UX * UY; |
942 | |
943 | |
944 | const bool IsNegative = (X < 0) ^ (Y < 0); |
945 | Result = IsNegative ? (0 - UResult) : UResult; |
946 | |
947 | |
948 | if (UX == 0 || UY == 0) |
949 | return false; |
950 | |
951 | |
952 | |
953 | |
954 | if (IsNegative) |
955 | return UX > (static_cast<U>(std::numeric_limits<T>::max()) + U(1)) / UY; |
956 | else |
957 | return UX > (static_cast<U>(std::numeric_limits<T>::max())) / UY; |
958 | } |
959 | |
960 | } |
961 | |
962 | #endif |